Computer Chess Club Archives


Search

Terms

Messages

Subject: SB correction rule in Swiss trn with too few participants

Author: Theo van der Storm

Date: 14:25:41 10/29/04

Go up one level in this thread


On October 24, 2004 at 17:16:55, Ed Schröder wrote:

>On October 24, 2004 at 14:11:06, Theo van der Storm wrote:
>
>>On October 24, 2004 at 13:03:15, Volker Richey wrote:
>>
>>>Round 1:  Kallisto - Diep     0 - 1
>>>Round 3:  Nexus - IsiChess    1 - 0
>>>Round 3:  Kallisto - ProDeo   0 - 1
>>>Round 8:  Diep - Goldbar      1 - 0
>>>Round 11: IsiChess - ProDeo   0 - 1
>>>Round 11: Nexus - XiniX       1 - 0
>>>
>>>1. Diep    38,00
>>>2. Nexus   37,75
>>>3. ProDeo  37,50
>>>
>>>Volker
>>
>>The games you mentioned appear to be correct,
>>but in the SB point it seems you erroneously subtracted
>>three Neurosis games worth 2 SB for each as well.
>>
>>So I wrote:
>>1. Diep    8,5   40.00
>>2. Nexus   8,5   39.75
>>3. Pro Deo 8.5   39.50
>>
>>The rule was elected by the participants in an earlier
>>tournament and published before the start of this tournament.
>>
>>Best,
>>Theo
>
>
>If the rules of 2004 would have been used in 2001 would Rebel have won then?
The rule hadn't been invented yet.

>http://www.computerschaak.nl/docc01.html
>
># Name              1 2 3 4 5 6 7 8 910111213141516 P  BU   SB   G
>1 Chess Tiger 14.6  X 1 1 1 1 0 ½ 1 . 1 ½ 1 1 . . . 9  70  57¾  11
>2 Rebel Century 4   0 X 1 1 ½ 1 1 ½ 1 1 1 . . . . 1 9  66  50¾  11
>
>I think the answer is yes, I remember that Rebel lost many points because it had
>to play EEC in the last round while it had a lead before the last round.

The answer is no, because numbers 11 and 12 together had 9.5 points
reducing Tiger's SB to 48.25
Numbers 9 and 16 together had 5.5 points reducing Rebel's SB to 45.25

>I am not complaining, rules are rules, but a play-off somehow sounds more fair
>than randomness. Just a consideration.
>
>Ed

If the game between the top-2 is not a draw, then that game will usually be
dominant for their SB and with good reason. It's certainly not random.
In 2004 we have a top-3, so it gets more complicated.

Still, the rule has good statistical properties and gets rid of the effect
of bad pairings. I consider a blitz play-off more random, or even worse
than that: more a reflection of the quality of the operators in stead
of the programs.

I respect your opinion, but as you will understand I disagree.
I hope you appreciate the fact that we asked the Oct. 2003
participants for their opinion and implemented just that.

Theo

PS: Good Luck with Pro Deo next time!



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.